As an important technology in the film industry, Multispectral Lighting Reproduction is used widely to provide real and vivid images for audiences. The overlapping shadows cause the redundant part of the film images which lead to uncoordinated visual effects. This problem is confusing the film industry. To avoid this problem, light sources have to irradiate from a far distance inside the gigantic building, meanwhile a high cost of equipment is required. Hence, it is important to propose a better method which can shorten the illuminative distances and save a huge amount of cost while maintaining the good accuracy of lighting reproduction. In this paper, we propose a novel system to solve the overlapping shadows problem, which can remarkably reduce the power and hardware costs in multispectral lighting reproduction system. The system consists of two workflows. First, the non-negative least squares solution is used to solve the illuminance coefficient. The accurate light is achieved by the illuminance coefficient. Second, images are categorized as the full shadow, overlapping shadows and non-shadow by different luminance and color features. Finally, overlapping shadows are removed by the inpainting algorithm. Experimental results show that our method can effectively detect the overlapping shadows and images inpainting of the dataset.